John Myles White

I’m happy to announce that I’ve started writing a new book for O’Reilly, which will focus on teaching readers how to use Multi-Armed Bandit Algorithms to build better websites.

At MSR this week, we had two very good talks on algorithmic methods for tuning the hyperparameters of machine learning models.

Introduction Link to heading In the third installment of my series of criticisms of NHST, I focused on the notion that a p-value is nothing more than a one-dimensional representation of a two-dimensional space in which (1) the measured size of an effect a…

Update 10/30/2013: Since this post was written, Julia has acquired a large body of optimization tools, which have been grouped under the heading of JuliaOpt.

On July 25th, I’ll be presenting at the Seattle R Meetup about implementing Bayesian nonparametrics in R.

Many readers of this blog will know that I’m a big fan of Bayesian methods, in large part because automated inference tools like JAGS allow modelers to focus on the types of structure they want to extract from data rather than worry about the algorithmic…

[Note to the Reader: This is a much rougher piece than the previous pieces because the argument is more complex.

Introduction Link to heading Continuing on with my series on the weaknesses of NHST, I’d like to focus on an issue that’s not specific to NHST, but rather one that’s relevant to all quantitative analysis: the destruction caused by an inappropriate reducti…

Introduction Link to heading This is my second post in a series describing the weaknesses of the NHST paradigm.

Introduction Link to heading Over my years as a graduate student, I have built up a long list of complaints about the use of Null Hypothesis Significance Testing (NHST) in the empirical sciences.

Introduction Link to heading For me, Hadley Wickham’s reshape and plyr packages are invaluable because they encapsulate omnipresent design patterns in statistical computing: reshape handles switching between the different possible representations of the s…

One major benefit of spending my time recently adding statistical functionality to Julia is that I’ve learned a lot about the inner guts of algorithmic null hypothesis significance testing.

While I should confess upfront that I’ve always had a weaker command of the details of floating point arithmetic than I feel I ought to have, this sort of thing still blows my mind when I stumble upon it.

While exploring the Julia manual recently, I realized that it might be helpful to put the basic vocabularies of Julia and R side-by-side for easy comparison.

Building Optimization Functions for Julia Link to heading In hopes of adding enough statistical functionality to Julia to make it usable for my day-to-day modeling projects, I’ve written a very basic implementation of the simulated annealing (SA) algorith…

Julia is a new language for scientific computing that is winning praise from a slew of very smart people, including Harlan Harris, Chris Fonnesbeck, Douglas Bates, Vince Buffalo and Shane Conway.

If you’re subscribed to this blog, you’ve surely noticed the very long hiatus I’ve taken from writing over the last six months.

Introduction Link to heading I’ve recently been working with a couple of large, extremely sparse data sets in R.

Introduction Link to heading This semester I’m TA’ing a course on the Psychology of Music taught by Phil Johnson-Laird.

Yesterday I posted a very simple math puzzle to Twitter that I found in Jonathan Baron’s book, Thinking and Deciding.

Yesterday the Princeton machine learning reading group went through a paper by Tukey on “Some graphic and semigraphic displays”.

The news below was recently reported on the ProjectTemplate mailing list.

Recently, I’ve been fitting some models from the behavioral economics literature to choice data.

In this, Chomsky is in complete agreement with O’Reilly.

This morning I tried to completely update my R installation.

[UPDATE 3/28/2011: Fixed an enormous bug in the R code.]

I may have deleted a genuine comment by accident today while cleaning out the spam queue.

I like exploring the behavior of functions of a single variable using the curve() function in R.

Introduction Link to heading Neil Kodner recently got me interested again in analyzing Canabalt scores statistically by writing a great post in which he compared the average scores across iOS devices.

The kind people at Packt Publishing recently asked me to review one of their newest R books: the R Graphs Cookbook.

The Bayesian-Frequentist debate reflects two different attitudes to the process of doing science, both quite legitimate.

But modeling devices that make sense for an unbiased decisionmaker may not make sense for a biased one.

Today I needed to quickly reverse the order of rows in a LaTeX table.

We are constantly exploited by the tools meant to foil our exploitation.

Every academic field seems to develop a set of generic insults based on their intellectual toolkit.

I’ve just uploaded a new binary of ProjectTemplate to GitHub.

New York’s annual marathon took place yesterday.

To quote from the preface to the first edition in Jeffreys (1961): ‘It is sometimes considered a paradox that the answer depends not only on the observations but on the question; it should be a platitude.’

As a quick note, here are two R packages that were mentioned to me recently and that look promising: reldist and mixtools.

[UPDATE: As Will points out in the comments, this isn’t really the EM algorithm.

The R Recommendation Engine contest is now live on Kaggle.

On Dataists, a new collaborative blog for data hackers that I’m contributing to, we’ve just announced a data contest that’s custom made for R users.

I’ve just released the newest version of ProjectTemplate.

Other than our culture’s implicit association between lies, damned lies and statistics, I think no idea has stifled the growth of statistical literacy as much as the endless repetition of the words correlation is not causation.

On the other hand, taken as a whole, decision theory seems to have about the same connection to real decisions as war games played on a table do to real wars.

I’ve just released two new packages for R: log4r and SortableHTMLTables. log4r is a minimal logging utility for R that’s inspired by the log4j family of logging tools.

Introduction Link to heading Because R is, in part, a functional programming language, the ‘base’ package contains several higher order functions.

Here’s a nice snippet from a 2009 article by Kass that I read yesterday: According to my understanding, laid out above, statistical pragmatism has two main features: it is eclectic and it emphasizes the assumptions that connect statistical models with obs…

We use frameworks like Ruby on Rails or ProjectTemplate to minimize the time we spend on irrelevant details.

I’m currently reading Luce’s “Response Times”.